Silvia Pineda-Munoz and John Alroy of Macquarie University in Sydney, Australia have looked carefully at all the terrestrial mammals. Their aim was to see how varied a carnivorous diet or a herbivorous diet could be so that the requirements of an ecosystem could be more accurately estimated. Our present-day placentals and even the marsupials are extraordinarily diverse. As far as niches go, they occupy most of the ground-based, the aerial and various aquatic ecologies. And they eat almost anything they can get their molars into.
The results of the survey leaves us with 7 clusters of dietary habit:
GRANIVORY (eating grasses)
GUMNIVORY (sap extracting,) AND
FRUGIVORY (eating fruit.)
The carnivores form a group that eat large or small prey, while they may, like tigers, consume some fruit, berries or grass at times. It is this deviation from the expected that the authors are measuring here, in order to asses any ecosystem better. The only fully gumnivorous species recognised is one of the African bush-babies, Euoticus elegantulus. Likewise, the fungivores are represented by very few species such as some chipmunks.
As might be expected, the biggest problems were the generalists, if you like, the omnivores, who eat almost anything available. In fact these omnivore generalists must be divided into very different adaptive groups. We could fit easily into that category! The rodents fit into almost every category, exploiting all major food resources, and especially exploit the generalist niches with 90% of their members.
Rodents apart, very few other mammals present a pure dietary profile either. With exact amounts of different foods impossible to judge quantitatively, even for living species, the 7 classes proposed leave what sees the only feasible informing scheme. Only the most frequently consumed food resource is used, so we are far from classifying any niche. We do have the option of specialist (more than 50% of diet on one resource) or generalist (no resource reaches the 50% level). Any resource taking up 20-50% of the diet would also be mentioned. This gives us a clue about differences within genera and helps with fossil classifications, if the information on possible diet is available.
Mammals emerge as a highly-specialised group. Their feeding habits diversify when it seems parasite and predator avoidance or competition is involved. Ecological research should benefit from this scheme, as it helps construct past and present food webs. Evolving the scheme into a more and more useful database will be a function of those using it many species are yet to be fully described in terms of their diet. This will help a lot towards that aim. You can read the detail of the research in the Proc.Roy.Soc.B's - Dietary characterization of terrestrial mammals.